|Home | About | Journals | Submit | Contact Us | Français|
The OPTN classifies high infectious risk donors (HRDs) based on criteria originally intended to identify people at risk for HIV infection. These donors are sometimes referred to as "CDC high risk donors" in reference to the CDC-published guidelines adopted by the OPTN. However, these guidelines are also being used to identify deceased donors at increased risk of window period (WP) hepatitis C virus (HCV) infection, although not designed for this purpose. The actual risk of WP HCV infection in HRDs is unknown.We performed a systematic review of 3,476 abstracts and identified 37 eligible estimates of HCV incidence in HRD populations in the United States/Canada. Pooled HCV incidence was derived and used to estimate the risk of WP infection for each HRD category. Risks ranged from 0.26–300.6 per 10,000 donors based on WP for ELISA and 0.027–32.4 based on nucleic acid testing (NAT). Injection drug users were at highest risk (32.4 per 10,000 donors by NAT WP), followed by commercial sex workersand donors exhibiting high risk sexual behavior (12.3:10,000),men who have sex with men (3.5:10,000), incarcerated donors (0.8:10,000), donors exposed to HIV infected blood (0.4:10,000), and hemophiliacs (0.027:10,000). NAT reduced WP risk by approximately 10-fold in each category.
In 1985(1), and later updated in 1994 (2), the Public Health Service (PHS) developedguidelines to identify persons at increased risk for HIV infection. These guidelines were published by the Centers for Disease Control and Prevention (CDC) and adopted by the OPTN to identify deceased organ donors at increased risk of infectious transmission. Currently, any donor falling into one or more of the 7 specified behavioral categories (Table 1) is classified as a high risk donor (HRD) and subject to additional OPTN policies. Despite significant controversy surrounding their use, approximately 9% of donors where at least one organ is recovered are categorized as HRDs (3).
While the originally intended purpose of the HRD guidelines was to identify those at risk of prevalent infection, in practice this is not a relevant concern. UNOS mandates HIV and HCV antibody testing for all deceased donors and, as such, prevalent infections are detected (4, 5). However, no serologic test can detect infections that occurred very recently, so theconcern isincident infection occurring during the window period (WP), the time between acquisition of infection and serologic detectability. This scenario will likely result in a false-negative test result and transmission to the recipient(6–8). As such, the HRD criteria are instead used as surrogate criteria to identify persons at risk of recently acquired infection
The PHS/CDC guidelines have not only been extrapolated in practice from prevalent HIV to WP incident HIV infection, but they have also been used as a default proxy for identifying deceased donors at risk for WP incident infections with hepatitis C virus (HCV)(9).While HIV and HCV share several modes of transmission, their epidemiology is not identical. Both are blood-borne illness that can bespread parenterally(10, 11); however HCV transmission is thought to be more efficient by this route than HIV infection. Sexual transmission is considered one of the primary drivers of the HIV epidemic(12); in contrast, sexual transmission of HCV is thought to be extremely inefficient, if at all relevant(13).Several studies of HCV discordant couples found little to no evidence of sexual transmission(14, 15). Having multiple sexual partners may increase the risk of HCV infection(16, 17); however, it is unclear whether this is a real effect or the result of other confounding risk behaviors(18).Three of the HRD categories are based on sexual behaviors (MSM, CSW, and high risk sexual behavior). It is unclear whether these categories are truly indicative of an increased risk for HCV.
An additional problem with expanding guidelines intended for HIV to HCV is that the two diseases have distinct clinical courses with different implications for serologic testing. While antibodies to HIV are typically produced within 3 weeks of infection (19), HCV antibody formation does not occur for 8–12 weeks (20), making the WP between acquisition of infection and serologic detectability by Enzyme-Linked Immuno- Assay (ELISA), an antibody-based method, much longer for HCV (approximately 66 days for HCV compared with 22 for HIV). Nucleic acid testing (NAT) is an alternative method based on detection of viral particles that shortens the WP to approximately 1 week (21). While antibody testing is mandated for all donors (22), the decision to use NAT is left to the individual Organ Procurement Organization (OPO). A survey of OPOs performed in 2008 found that 48.3% performed HCV NAT for all donors and an additional 20.7% performed it under certain circumstances (i.e. for HRDs only, when requested by the transplant center) (23). NAT is not universally used because it is more expensive, time consuming, and thought to have a higher rate of false positives compared to ELISA (9, 24). However, given that HCV NAT decreases the WP by several months, the benefits for HCV detection may outweigh the risks in a more pronounced way than with HIV.
We hypothesized that the risk of HCV WP infection would be higher than the risk of HIV for some categories and significantly lower for others, and that quantifying these risks would be essential for informed clinical decision-making with regards to HRDs. The goals of our study were to (1) estimate the incidence and variance of HCV infection within each category of HRD behavior, and (2) estimate the risk of HCV WP infection, by ELISA and NAT, within each category of HRD behavior.
We performed a PubMed search on November 27th, 2008 for studies of HIV or HCV incidence or prevalence (see appendix for details of search). Studies were eligible for inclusion in our systematic review if they presented an original estimate of HIV or HCV prevalence or incidence in a population located in the United States or Canada on or after January 1, 1995. We chose this cut-off for two major reasons: (1) accurate screening tests for HCV were not developed until the early 1990s and screening of the blood supply dramatically changed the epidemiology of HCV, particularly among hemophiliacs (25); and (2) the dynamics of HIV transmission, and thus HIV/HCV coinfection, likely changed with the introduction of highly-active antiretroviral therapy. Two independent reviewers and two adjudicators performed a systematic review for manuscripts meeting inclusion criteria and mined references of a 20% sample of eligible studies to identify any studies that might have been missed (26). Since incidence studies require significant resources for the identification and follow-up of large numbers of individuals, we hypothesized that many studies would have NIH funding. As such, we searched the NIH grant database for keywords"HIV" or "HCV" and "Incidence" funded after 1995 in order to identify any studies that might have been missed by other methods..
The goal of this meta-analysis was to estimate HCV incidence in HRD populations; as such only a subset of studies included in the systematic review were eligible for our meta-analysis. Eligible studies reported an estimate of HCV incidence in one of the 7 HRD populations outlined in Table 1. Studies reporting HCV prevalence estimates were included for HRD behavioral categories for which multiple incidence studies were unavailable Approximately 80–85% of persons exposed to HCV develop chronic infection with persistent HCV-RNA positivity while the remaining 15–20% eventually clear the virus(27). However, even those with eventual clearance were likely infectious during the acute phase, the point of interest when the concern is WP infection.As such,for consistency,it was decideda priori that only studies using HCV antibody testing only or a combination of HCV antibody and RNA (where everyone was tested with both and considered infected if positive by either method) would be included; fortunately, there were no studies in HRD populations that used only RNA without antibody testing, so no studies had to be excluded. Estimates lacking a denominator (for example, reported HCV cases over the total population) were also excluded.
Data from each eligible study were abstracted by at least 2 independent reviewers and disagreements were adjudicated as previously described (26). The following data were abstracted from each article: recruitment dates, county, state, city, and location where recruitment took place, sampling method (convenience, target, random sample, chain/referral sampling), inclusion criteria, testing method, number of patients approached, number eligible, number tested, number positive by HCV antibody and number positive by HCV RNA. We abstracted the overall HCV incidence and prevalence for each article, in addition to risk-stratified sub-estimates if the risk factor was one of the HRD criteria. For incidence studies, the number of seronegative patients eligible for follow-up, the number tested at follow-up, the number of seroconversions, incidence rate, and total number of person-years at risk were abstracted, back-calculated using other data from the manuscript, or obtained directly from the study authors.
Any study of HCV incidence among persons demonstrating one of the HRD behavioral risk factors was eligible for inclusion in our meta-analysis. When we were unable to find multiple incidence studies in a behavioral category, we included all prevalence studies in this category and estimated incidence from prevalence as described below. Each HCV incidence estimate was classified as falling into one of the categories, falling into multiple categories, or other. All studies in the same behavioral category that took place in the same geographic location were flagged and re-examined to ensure that the estimate were not derived from the same cohort and could be combined without fear of counting the same individuals multiple times. Pooled incidence estimates were obtained by summing the person-years and number of HCV seroconversions for each study within categories of HRD behavior. Persons were considered to have seroconverted if they became HCV antibody or HCV-RNA positive over the course of the study. Poisson exact 95% confidence intervals were calculated for each pooled incidence estimate using Stata 11/MP (College Station, TX).
Incidence studies are challenging and resource intensive, requiring recruitment, follow-up, and serologictesting of large cohorts over long periods of time. Furthermore, three of the HRD behavioral categories involve sexual risk factors, widely considered to be an inefficient mode of HCV transmission. As such, there have been few incidence studies of HCV specifically aimed at populations with sexual risk factors. To account for this, data were abstracted from HCV prevalence studies in each category, and methods for estimating the incidence of a disease in a given population based on its prevalence were used when insufficient incidence studies were found(28).Briefly, we compared the ratio of incidence to prevalence in an HRD population where both were known, in order to solve for the unknown incidence value using the pooled prevalence estimate in the HRD population of interest:
Upper and lower bounds for the unknown incidence were computed by substituting the upper and lower bounds of the Poisson Exact 95% CI computed for HCV incidence in IDUs.
Pooled incidence estimates from the meta-analysis were then used to calculate a per-day incidence rate (IR) and then combined with WP duration to calculate the probability of a WP infection using iterative conditional probabilities as shown in the following equation(26):
Upper and lower bounds on the probability of a WP infection were calculated by using the upper and lower bounds of the Poisson exact 95% CIs around the pooled incidence rates for the pooled incidence value in the equation above.
Since very few studies of hemophiliacs were available since recent significant improvements in blood screening (29), estimates in hemophiliacs were determined based on studies of HCV incidence in blood donors described previously in more detail(26). Briefly, we used the WP of the serologic tests used to screen blood donors and HCV incidence in this population to calculate the residual risk of HCV in the blood supply:
We then used this residual risk to calculate the probability that a hemophiliac might contract HCV, making the assumption that 1 unit of blood per day was transfused for the entire duration of the HCV NAT WP:
This category, taken literally from the HRD guidelines (based on the mechanism that percutaneous exposure to HIV infected blood might transmit HIV), makes little sense in the context of HCV (since percutaneous exposure to HIV infected blood will not, by mechanism, transmit HCV). However, to be most conservative, we felt the best approach to risk estimation in this category was to combine estimates of HCV infectious risk per percutaneous exposure and estimates of HCV prevalence among persons infected with HIV. A recent multicenter study in Italy followed persons percutaneously exposed to HCV infected blood over 55 hospitals and calculated the risk of HCV transmission per exposure involving blood infected with HCV only as well as blood coinfected with HIV and HCV. They found that the rate of HCV transmission was over twice as high when the blood was HIV and HCV coinfected as when the blood was infected with HCV only (0.35% versus 0.85%). As such we used an estimate of0.85% per exposure in our calculations. To estimate the percent of HIV infected blood coinfected with HCV, we compiled prevalence studies of HCV infection among persons with HIV. We used the per needlestick estimate for coinfected blood combined with the probability that HIV infected blood is coinfected with HCV to calculate the risk of WP infection, assuming only one exposure event occurred, and there was equal probability of the event occurring on any day in the year prior to death.
After screening, 337 articles were eligible for inclusion at the full-text level (Figure 1) and 103 were eligible for data abstraction. For the meta-analysis, these were further narrowed to 37 unique estimates among populations meeting HRD behavioral criteria.
Six eligible prevalence studies were found in this category, for a total pooled sample size of 1341 participants (Table 3)(30–35). The incidence was estimatedusing established techniques(28)based on prevalence among IDUs as the reference category (32–65). Incidence was calculated to be 1.8 per 100 person-years (range 1.7–2.0, Table 2). Per 10,000 donors, the estimated risk of WP infection was 32.5 (range 30.7–36.1) with ELISA and 3.5 (range 3.3–3.8) with NAT.
Nine studies of HCV incidence in IDUs were identified for a total pooled sample size of 1955 participants(Table 4)(43, 45, 47, 49, 52, 54, 66–68). Incidence rates ranged widely from 0.68 to 35.9 per 100 person-years; however, when limiting to studies that only recruited persons who reported injection in the preceding 6 months, the range was 10 to 35.9 per 100 person-years. Pooled HCV incidence among IDUs was 16.9 per 100 person-years (range 15.5–18.4, Table 2). Per 10,000 donors, the estimated risk of WP infection was 300.6 (range 276.1–326.8) with ELISA and 32.4 (range 29.7–35.3) with NAT.
Results from 23,952, 635 individual blood donations from Canada and the United States were included in the pooled estimate (Table 5)(69, 70). HCV incidence in Canada was 0.00163 per 100 person-years and in the United States was 0.00189 per 100 person-years. The pooled HCV incidence among blood donors was estimated to be 0.0018 per 100 person-years (range 0.22–0.32, Table 2). Per 10,000 donors, the estimated risk of WP infection was 0.26 (range 0.22–0.32) with ELISA and 0.027 (0.023–0.034) with NAT.
Seven prevalence studiesof CSWs were identified, for a total pooled sample size of 678 participants (Table 6)(31, 34, 35, 39, 51, 60, 63). Using methods described above (see MSM category above), incidence was estimated to be 6.4 per 100 person-years (range 5.9–7.0, Table 2). Per 10,000 donors, the estimated risk of HCV WP infection was 114.9 (range 105.9–125.6) with ELISAand 12.3 (range 11.3–13.4) with NAT.
Eighteligible prevalence studies were identified, with a total pooled sample size of 1361 participants (Table 7)(31, 33, 35, 39–41, 51, 60). Incidence was estimated to be 6.4 per 100 person-years (range 5.8–6.9, Table 2). Per 10,000 donors, the estimated risk of HCV WP infection was 114.9 (range 104.2–123.8) with ELISA and 12.3 (range 11.1–13.2) with NAT.
A recent large 55-hospital prospective cohort study of workers percutaneously exposed to HCV infected blood found a per exposure risk of 0.85% when the blood was also coinfected with HIV (Table 8a)(71). To estimate the probability that HIV infected blood was coinfected with HCV, we compiled prevalence studies of HCV among HIV positive persons, identifying 4 studies for a total pooled sample size of 6736 participants (Table 8b)(39, 49, 72, 73). Per 10,000 donors, the estimated risk of HCV WP infection was 4.0 (range 0.9–11.1) with ELISAand 0.4 (range 0.09–1.2) with NAT(Table 2).
We were only able to identify one intra-prison incidence study of HCV incidence with a total sample size of 337 participants (Table 9)(74). Incidence was estimated to be 0.4 per 100 person-years (95% CI 0.04–1.3, Table 2). Per 10,000 donors, the estimated risk was 7.2 (range 0.7–23.5) with ELISA and 0.8 (range 0.08–2.5) with NAT.
We found that the risk of HCV WP infection varied significantly across HRD behavioral categories and testing methods.Estimated WP risk of HCV per 10,000 donors ranged from 0.26–300.6based on WP for ELISA and from 0.027–32.4 based on WP for NAT. This is significantly higher than the estimated WP risk of HIV infection which ranged from 0.04–12.9 per 10,000 donors in a similarly conducted systematic review and meta-analysis(26). IDUs carried the highest risk of HCV WP infection (300.6 per 10,000 donors with ELISA and 32.4 with NAT), followed by commercial sex workers and donors engaging in high risk sexual behavior (114.9 per 10,000 donors with ELISA and 12.3 with NAT), MSMs (32.5 per 10,000 with ELISA and 3.5 with NAT), incarcerated donors (7.2 per 10,000 donors with ELISA and 0.8 with NAT), donors exposed to infected blood (4.0 per 10,000 donors with ELISA and 0.4 with NAT), and hemophiliacs (0.26 per 10,000 with ELISA and 0.027 with NAT).Relative order of risk differed somewhat from that for HIV WP infection, where IDUs also carried the highest risk of HIV WP infection but were followed by MSMs, CSWs, incarcerated donors, donors exposed to HIV through blood, donors engaging in high risk sexual behavior, and hemophiliacs. It is important to note that these estimates are only applicable to donors in the United States and Canada from which our data are drawn, and these estimates may be quite different in other parts of the world.
Previous studies have shown that the risk of sexual transmission of HCV is very low(14, 15). It is thought that the risks might increase for persons with multiple sexual partners, STIs, and HIV infection but this has not been shown definitively(16–18). Commercial sex workers, persons engaging in high risk sexual behavior, and MSMs were among the highest risk categories for HCV WP infection in our analysis. While this might be partially explained by sexual risk factors, it is possible that the higher incidence in these populations is reflective of confounding resulting from high likelihood of exhibiting other high risk behaviors, such as injection drug use, shown to result in very efficient HCV transmission.
Our study has several limitations. Incidence and prevalence studies often target higher risk individuals to ensure a sufficient number of cases to examine outcomes of interest. As such our results are not necessarily generalizable to the underlying populations, and may be overestimates of the true WP risk.Another issue is the potential for overlap between categories. While we excluded estimates in persons with multiple risk factors, most studies did not measure all HRD behaviors and as such incidence in one category might be explained not by the risk of the behavior itself but by its correlation with another risky behavior. This is especially true for commercial sex workers who have been previously shown to have high rates of injection drug use.
Our study is the first to systematically report the risk of HCV WP infection in donors meeting the PHS/CDC high risk criteria, and we found a fairly significant risk of WP infection, particularly in the IDU category. HCV-specific antibodies are typically not detectable until 2 months or longer after acquisition of infection; as such the risk of WP infection in populations is quite high when ELISA is used. NAT, with a WP of approximately 7 days, reduces the risk of WP HCV infection by an order of magnitude compared to ELISA, from approximately 3 in 100 to 3 in 1000 among IDUs. Our findings suggest that the use of these categories to identify persons at risk of HCV infection is not unreasonable as risk of HCV WP infection is very high, particularly for injection drug users. However, for some categories (hemophiliacs and persons exposed to HIV (+) blood), the risk was minimal. Furthermore, because the guidelines were not specifically developed for HCV, there may be other behavioral risk factors not included in the criteria that place a person a high risk of acquiring HCV that are not currently captured.Previous studies have suggested that tattooing, body piercing, and intranasal cocaine use might be important routes of HCV transmission (75, 76), as such these risk factors should be evaluated as potential additional categories in any future revisions of these guidelines.
In conclusion, we hope that these data, combined with those of HIV WP infection risk among patients in the same behavioral categories, will help clinical decision-making and counseling with regards to HRD organs.
This publication was supported by Grant Number UL1 RR 025005 from the National Center for Research Resources (NCRR), a component of the National Institutes of Health (NIH), and NIH Roadmap for Medical Research. It was also supported by Grant Number R21DK089456 from the National Institute Of Diabetes And Digestive And Kidney Diseases (NIDDK). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIDDK, the NIH, or the NCRR.
The authors have no conflict of interest to disclose. This study was not funded in any way by a commercial organization.